Extracting soil moisture information from ERS-1 imagery

Abstract

Active microwave remote sensing techniques have been shown to be effective in measuring moisture levels at the soil surface. As synthetic aperture radar (SAR) data from satellites increasingly becomes available, the potential exists to repetitively monitor soil moisture conditions over large areas. Here, multitemporal ERS-1 SAR imagery were integrated with digital terrain data, vegetation measurements, and precipitation records to: (1) quantify and reduce the effect of variable terrain on SAR image digital number (DN) values, (2) isolate the amount of soil-contributed backscatter (σ_SOIL⁰) from the total backscatter coefficient (σ_TOT⁰), and (3) qualitatively examine the relationship between σ_SOIL⁰ and surface soil moisture conditions over tallgrass prairie subjected to varying treatments of prescribed burning. Local incidence angle (LIA) was found to have the strongest influence on SAR image backscatter values (r = 0.341). An empirically derived correction function based on the least squares estimate of σ_TOT⁰ and LIA data pairs improved image quality by more than 11%. A simple cloud model is then used to calculate the amount of radar energy backscattered by the vegetation canopy (σ_VEG⁰) in both burned and unburned watersheds. Subtracting σ_VEG⁰ from σ_TOT⁰ provides the estimate of σ_SOIL⁰. Unburned watershed σ_SOIL⁰ was significantly lower than that of burned watersheds, due to the higher density of scattering particles causing elevated σ_VEG⁰ values in these areas. At watershed and larger scales, changes in σ_SOIL⁰ reflected temporal changes in surface soil moisture levels inferred by local precipitation measurements and water budget consideration. © 1998 John Wiley & Sons, Inc.